Parameter Sharing
A mechanism in which multiple sequences or nodes share the same set of model parameters, providing an inductive bias and a regularization effect that reduces the risk of overfitting.
Timeline (last 90 days)
In power load forecasting, the GNN simultaneously performs cross-series parameter sharing and neighbor-series reading, and the model's improvement over non-graph models may stem mainly from parameter sharing rather than neighbor reading.
On the feeder data, parameter sharing recovers about two-thirds of the error gap between node-specific models and the GNN (relative to the regularized baseline).
All Facts (2)
In power load forecasting, the GNN simultaneously performs cross-series parameter sharing and neighbor-series reading, and the model's improvement over non-graph models may stem mainly from parameter sharing rather than neighbor reading.
50%UnverifiedOn the feeder data, parameter sharing recovers about two-thirds of the error gap between node-specific models and the GNN (relative to the regularized baseline).
50%